Article Text

Download PDFPDF

Procedural sedation and analgesia in a large UK Emergency Department: factors associated with complications
  1. Keith G Jacques1,
  2. Alistair Dewar2,
  3. Alasdair Gray2,
  4. Dean Kerslake2,
  5. Alexis Leal2,
  6. Fiona Lees3
  1. 1Emergency Department, Stirling Royal Infirmary, Stirling, Scotland, UK
  2. 2Emergency Department, Royal Infirmary of Edinburgh, Edinburgh, Scotland, UK
  3. 3NHS National Services Scotland, Paisley, Scotland
  1. Correspondence to Dr Keith G Jacques, Consultant in Emergency Medicine, Stirling Royal Infirmary, Livilands, Stirling, FK8 2AU, UK; keith.jacques{at}


Background The aim of this study was to investigate the delivery of procedural sedation and analgesia (PSA) in an urban teaching hospital Emergency Department (ED) over a 2-year period, and identify factors associated with complications.

Methods Consecutive ED patients requiring PSA were prospectively enrolled onto the Registry of Emergency Procedural Sedation. Data collected included: patient and procedural characteristics; process times; physiological parameters; agents; sedation depth; and composition of staff team and complications. Multivariable binary logistic regression was used to identify factors associated with an increased risk of complications.

Results 1402 patents required PSA during the study period. 1345 (95.9%) underwent orthopaedic manipulations. 597 (42.6%) received moderate sedation. 401 (28.6%) were sedated to deeper levels. Complications occurred in 49 (3.5%) cases. Deeper levels of sedation and the procedure occurring overnight were identified as risk factors for complications. Procedure type, patient age, grade of doctor and choice of drug were not found to be associated with an increased risk of complications.

Conclusions Procedural sedation and analgesia by Emergency Physicians is safe and effective; however, complications do happen. Complications are more likely at deeper levels of sedation and at night. Emergency Physicians must have the necessary skills and equipment to deal with such complications when they arise. EDs must be adequately staffed with trained clinicians 24 h a day to provide PSA.

  • Moderate sedation
  • deep sedation
  • analgesia
  • emergency medicine
  • anaesthesia
  • analgesia/pain control
  • management, risk management

Statistics from


Procedural sedation and analgesia (PSA) refers to the technique of administering sedatives or dissociative agents with analgesics to induce a state that allows the patient to tolerate unpleasant procedures while maintaining cardiorespiratory function.1

The ability to safely deliver PSA in a number of environments has been questioned.2 In the UK much of this concern has been based on historical reports of mortality and morbidity in endoscopy patients.3

There are a number of reasons to believe that PSA practised by Emergency Physicians should be seen as a distinct entity. Patients presenting to the Emergency Department (ED) requiring PSA are, by definition, unprepared. Many will be unfasted, and there is limited opportunity to correct any physiological derangement. These factors potentially increase the risk of untoward events in ED PSA. Conversely, the training Emergency Physicians receive in delivering care to emergency patients, and in advanced airway skills places them in a unique position to safely manage these potentially difficult patients.

It is suggested that verbal contact be maintained throughout to minimise risk to patients.2 This has been termed moderate sedation.4 When patients respond only to painful stimuli, this has been termed deep sedation. At this level, it is thought that patients may not protect their own airway and a level of monitoring and skill equivalent to general anaesthesia is required.2 There are few data defining the levels of patient risk at different depths of sedation in the ED.

Case series of PSA in the general or adult ED setting from the UK5 6 have reported complication rates of 1–16%, with variation depending on setting, agents and definition of adverse events. Although low complication rates and associated adverse outcomes may be reassuring, they present a challenge to research. In order to conduct research into complications arising during PSA, large numbers of patients are required.

By comparison to North America, there are a paucity of European data examining PSA practice and outcomes. In particular, reliable data about complication rates and those factors associated with complications are lacking. Documenting this information is vital in ensuring patient safety and improving patient care. The Registry of Emergency Procedural Sedation (REPS) was established to study ED PSA and improve clinical practice.

The aim of this study was to present the first 2 years data from REPS, and also to describe those factors that are associated with an increased risk of complications.


Study design

REPS is a prospective observational registry study collecting routine data on consecutive patients undergoing ED PSA. This study was discussed with the Southeast Scotland Research Ethics Service, which classified it as service evaluation and therefore no formal ethical approval was required.

Setting and participants

This study was conducted at an adult, principally urban, teaching hospital ED with 107 000 new patient attendances in 2007. All patients requiring PSA between 4 September 2006 and 3 September 2008 were consecutively enrolled onto REPS. Patients requiring sedation for other reasons, such as to control delirium were excluded.

Sedation procedures in this ED are covered by departmental guidelines. PSA is delivered in one of the resuscitation rooms with at least two doctors and one nurse present. All patients should receive supplemental oxygen.

New doctors to the department must initially deliver sedation under direct senior supervision until judged competent. At the time of the study there was no formal assessment of competence. Only doctors who have completed an approved anaesthetic placement may use propofol, etomidate or ketamine. Otherwise, no restrictions are placed on the choice of drugs.

Data collection

The clinicians treating the patient collected study data using a standardised data collection form. Data were collected at each stage of the patient's management. Data were collected on 1. patient characteristics: age, gender and diagnosis; 2. therapies including seniority and specialty of physician, choice of drug, maximum level of sedation; and 3. procedural outcomes including complications.

The department makes use of locum doctors as part of its clinical team. These doctors have normally previously worked in the department in a substantive training post. It was not recorded whether the doctor treating the patient was employed on that shift as a locum doctor. Locums were entered onto the database by the level that they were expected to operate at during that shift.

Depth of sedation was assessed by the treating doctor using a five-point scale adapted from the American Society of Anesthesiology ‘Continuum of Depth of Sedation: Definition of General Anesthesia and Levels of Sedation/Analgesia’ (table 1).7 The original American Society of Anesthesiology scale was adapted to include a score for patients who remained fully alert.

Table 1

Sedation Score adapted from American Society of Anaesthesiology ‘Continuum of Depth of Sedation Definition of General Anaesthesia and Levels of Sedation Analgesia’

Data collection forms were collected weekly and a list of all sedation procedures was generated for the week. A member of the REPS team checked this list against departmental administrative systems, including the charge nurses' log for the resuscitation rooms and the ED computerised patient administration system, to ensure that all eligible patients had been enrolled. Any missing data were collected by interviewing the clinicians involved and reviewing the patient notes. These interviews normally occurred within 2 weeks of the sedation procedure.

Complications were defined a priori as oxygen desaturation (Spo2<90%) at any time; hypotension (systolic blood pressure <90 mm Hg); vomit; aspiration; apnoea; or cardiac arrest. Those performing the procedure were also able to report any other events that they felt were a complication of the procedure. The case notes of all patients reported to have suffered a complication were individually reviewed.

Data were stored and managed using Microsoft Excel 2003 and all analyses were performed within SPSS V. 17.0.

Risk factor analysis

A list of variables identified as potential risk factors for the occurrence of complications were extracted from the dataset. This list was created before analysis of the dataset and was designed to include a range of clinically relevant factors. All continuous variables, except for age, were grouped into clinically meaningful categories. Due to small numbers of complications (ie, <5) within some subgroupings, it was necessary to regroup some factors to a higher level prior to formal statistical analyses.

Each factor was screened individually using logistic regression and those with a p value of ≤0.25 were made available for the final multivariable binary logistic regression analyses. Using the selected variables, successive models were constructed using backward elimination by considering the contribution of each factor and the significance of the variable within the model. Risk factors were eliminated until a model showing only significant factors (p<0.05) remained. The p value, OR and associated 95% CI are reported for each factor.


One-thousand four-hundred and twenty patients were enrolled onto REPS during the study period. Eighteen were excluded because the patient had not been sedated (n=14) or there was a duplicate entry on the database (n=4). The median age was 50 (range 13–101) and the male to female ratio was 1.2:1.

The characteristics of the study population and the complication rate for each factor considered are described in table 2.

Table 2

Characteristics of study population and complication rate for each factor studied. ‘Other’ procedures include chest drain insertion, DC cardioversion and other upper and lower limb manipulations

Fifty-one complications were noted in 49 (3.5%) patients (table 3). The most common complications were desaturation (n=20) and apnoea (n=8). Three patients suffered an arrhythmia: one patient in his/her 20s who suffered a brief supraventricular tachycardia after receiving cyclizine; and two patients in their 80s, one who experienced a 1 min period of atrial fibrillation, and another who had a brief period of bradycardia that resolved without treatment.

Table 3

Observed complications

Five patients encountered either laryngospasm or bronchospasm. These patients (age range 17–76) had all received either propofol or ketamine. None had been sedated with midazolam. Two patients suffered humeral fractures during shoulder reductions.

Of those patients who experienced a complication: 22 had their sedation reversed with naloxone and/or flumazenil. In those patients whose complication was attributable to their sedation rather than the procedure, there was no evidence that any suffered lasting morbidity as a result of their complication.

Most patients reached a sedation score of 3 (n=597, 42.6%) or 4 (n=370, 26.4%). Patients with a sedation score of 4 or 5 had a higher median age (61, IQR 34–77 and 60, IQR 35–76 respectively) than those with lower sedation scores (45, IQR 26–67 for sedation score 3 or lower).

The observed complication rate was highest in those patients requiring prosthetic hip reduction (n=16, 7.8%). The group of patients undergoing hip reductions were older (median 77, IQR 67–84) than the study population as a whole (median 50, IQR 28–71). The success rate for hip reductions (74.8%) was lower than for other procedures (96.1%). Hip reduction patients were also more likely to be sedated to sedation scores of 4 or 5 (58.7%) than the general study population (28.6%).

There was a wide variation in choice of drugs. The most commonly used sedative agents were midazolam (n=1011) and propofol (n=336). By contrast, ketamine (n=26) and etomidate (n=11) were used much more infrequently. Drug use was analysed by three different methods: by considering the number of agents used, by the choice of analgesic (fentanyl vs morphine) and by the choice of sedative (midazolam vs propofol). No relationship between drug regime and adverse events was observed in any of these analyses.

Risk factor analysis

The results of the initial univariable screening are shown in table 4. Four factors had a p value <0.25; age, procedure, maximum sedation score and time of procedure. Of these factors, only two were found to be statistically significant when entered into the multivariable model (table 5). Due to missing data, only 1270 cases, 45 (3.5%) with complications, were included in the multivariable analyses.

Table 4

Results of univariable screening showing all factors identified as potential risk factors and made available for binary logistic regression analyses

Table 5

Factors found to be associated with increased risk of complications following multivariable logistic regression analyses

In the final multivariable model, the odds of a complication occurring was higher among patients having their procedure in the late evening/night (20:00–07:59) (p=0.044, OR 1.87, 95% CI 1.02 to 3.43) when compared to those admitted during the day.

The odds of a complication occurring was higher among patients with a maximum sedation score of 4 (p=0.000, OR 4.55, 95% CI 2.34 to 8.88 or 5 (p=0.000, OR 14.23, 95% CI 5.02 to 40.32) when compared to patients who had a maximum sedation score of 3 or less.


Although strenuous efforts were made during this study to maximise data capture, some variables were incomplete (shown as ‘not known’ in table 2).

This study only records events that occurred in the ED. It is possible that complications arising after the time of discharge have been excluded.

The present results looking at choice of drug should be interpreted with caution. This was not a randomised study. It is likely that many clinicians choose their agents based on their familiarity with the drugs involved, the fasting status of the patient, the patient's clinical status and any pre-existing conditions, and the procedure to be performed. The reasoning behind these decisions cannot be deduced from this study and it was not possible to control fully for these factors. As a result, the present study should not be used as evidence of the safety of one agent over another.

Due to its complexity, the issue of fasting was not considered in this study. The authors plan to study this issue separately in a later study.

This is a pragmatic study, which relied on clinicians accurately recording their own practice. This will raise concerns that bias may have arisen and that complications have been under-reported.

Notwithstanding these factors, it is noted that the reported complication rate in this study is similar to previously reported series,5 suggesting that the figures presented are robust.


To the authors' knowledge, this is the largest study to date of adult ED PSA in Europe. The data suggest that PSA can be safely delivered to a wide spectrum of patients in unplanned and emergency settings.

The factor most strongly associated with a higher rate of complications in this study was deeper levels of sedation. There was also an increased rate of complications for patients undergoing their procedure during the late evening and at night.

These findings support the guidance that the level of staffing and monitoring required for a patient undergoing deep sedation is the same as that required for general anaesthesia. However, practitioners can be reassured that sedation techniques that maintain verbal contact throughout are very safe.

Although in an ideal world it might be prudent to only sedate patients to moderate sedation level, pre-determined sedation targets are hard to achieve.8 Furthermore, many practitioners may feel that this level of sedation is inappropriately light for many orthopaedic manipulations. It is essential that anyone undertaking PSA in the ED setting has the necessary skills and equipment to deal with any complications that may arise, including cardiorespiratory impairment and airway obstruction.

The authors are not aware of any other studies that have reported an increased risk of complications at night for ED PSA. It is impossible to conclude from the present data exactly why this may be the case. It was possible to control for some possible confounding factors such as age and seniority of doctor through the present study design. However, other factors such as intoxication and ED congestion were not considered. The role of disturbed sleep patterns on decision-making ability in junior doctors is well known9 and may have played a part.

The present finding that complications are more likely to occur at night has clear implications for staffing of EDs. The authors support the recommendation from the UK College of Emergency Medicine that all EDs should have a doctor trained and experienced in Emergency Medicine present 24 h a day.10

This study has stimulated much discussion in the authors' service as to what constitutes a complication. Subsequent to the present study commencing, the Quebec Guidelines for Terminology and Reporting Adverse Events in ED PSA in Children11 have been published. It is likely that these guidelines will also lead to a standardisation in definitions of adverse events in PSA in adults.

The Quebec Group decided that adverse events should only be reported if they lead to a rescue action. The present study adopted a much broader approach, by including not just a pre-determined list of complications, but also by allowing clinicians to report any event that they felt constituted a complication. It is likely that had the Quebec Guidelines been used for the present study, then the rate of complications would have been even lower.

Whether events that occur during PSA in the ED are deemed adverse or not, the authors believe it is vital that open disclosure and discussion takes place to ensure continuous improvement in patient treatment and safety.

PSA can be safely delivered in the ED, but adequate training, staffing and equipment are essential.


We wish to thank the staff of the Royal Infirmary of Edinburgh Emergency Department for their help in completing this study. We also wish to thank STAG (Scottish Trauma Audit Group) for its support and advice in the early stages of this project.



  • Competing interests None.

  • Ethics approval This study was discussed with the Southeast Scotland Research Ethics Service, which classified it as service evaluation and therefore no formal ethical approval was required.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles

  • Primary survey
    Malcolm Woollard